Modeling Uncertainty in Latent Class Membership: A Case Study in Criminology

• DOI: 10.2307/2669989
• Publication Date: 2006-05-09T17:10:18Z
• Authors: Kathryn Roeder; Kevin G. Lynch; Daniel S. Nagin
• Keywords: Trait; Life course approach; Local independence

Abstract Social scientists are commonly interested in relating a latent trait (e.g., criminal tendency) to measurable individual covariates poor parenting) understand what defines or perhaps causes the trait. In this article we develop an efficient and convenient method for answering such questions. The basic model presumes that two types of variables have been measured: Response (possibly longitudinal) partially determine class membership, risk factors wish relate these variables. assumes observable conditionally independent, given variable. We use mixture joint distribution observables. apply longitudinal dataset assembled as part Cambridge Study Delinquent Development test fundamental theory development. This holds crime is committed by distinct groups within population: Adolescent-limited offenders life-course-persistent offenders. As labels suggest, distinguished longevity their offending careers. also predicts disproportionately comprised individuals born with neurological deficits reared caregivers without skills resources effectively socialize difficult child. Key Words: Classification errorLatent analysisMixture models